spacepaste

Paste details

reply | raw

  1.  
  2. #
  3. #
  4. # Mark Grandi - Oct 14, 2014
  5. #
  8. import argparse, sys, zlib, pprint
  10. HEADER="B+Tree Graph Index 2"
  11. NODEREF = "node_ref_lists="
  12. KEYELEMENTS = "key_elements="
  13. LEN = "len="
  14. ROWLENGTH = "row_lengths="
  16. def printOutIndexKeys(args):
  17. '''print out bzr index keys
  18. @param args - the namespace object we get from argparse.parse_args()
  19. '''
  21. import pdb;pdb.set_trace()
  23. def readUntilNewline(fileObj):
  25. ba = bytearray()
  27. while True:
  28. tmp = fileObj.read(1)
  29. if tmp == "\n".encode("utf-8"):
  30. return ba.decode("utf-8")
  31. else:
  33. ba.append(ord(tmp))
  35. def testString(fileObj, expectedString):
  36. test = fileObj.read(len(expectedString))
  37. if not test.decode("utf-8") == expectedString:
  38. sys.exit("String incorrect, expected {0} but got {1}".format(expectedString, test.decode("utf-8")))
  41. for iterFile in args.infileList:
  42. testString(iterFile, HEADER)
  44. testString(iterFile, "\n")
  46. testString(iterFile, NODEREF)
  48. noderef_val = int(readUntilNewline(iterFile))
  50. testString(iterFile, KEYELEMENTS)
  51. keyelements_val = int(readUntilNewline(iterFile))
  53. testString(iterFile, LEN)
  54. len_val = int(readUntilNewline(iterFile))
  56. testString(iterFile, ROWLENGTH)
  57. rowlength_val = readUntilNewline(iterFile)
  59. print("noderef: {0}, keyelements: {1}, len: {2}, rowlength: {3}".format(noderef_val, keyelements_val, len_val, rowlength_val))
  62. start = iterFile.tell()
  63. finish = 4096
  64. numLeaf = 0
  65. numInternal = 0
  66. totalKeysInInternal = 0
  68. while True:
  70. print("######################################")
  71. uncomp = iterFile.read(finish-start)
  73. if not uncomp:
  74. print("eof")
  75. break
  77. print("read from {0} to {1}".format(start, iterFile.tell()))
  79. decomp = zlib.decompress(uncomp)
  81. splitList = decomp.split(b'\n')
  82. if splitList[0].decode("utf-8") == "type=internal":
  83. numInternal += 1
  85. totalKeysInInternal += len(splitList) - 3 # remove type=, offset= and empty string at end
  86. if splitList[0].decode("utf-8") == "type=leaf":
  87. numLeaf += 1
  89. pprint.pprint(splitList)
  91. start = iterFile.tell()
  92. finish += 4096
  94. print("done, numLeaf: {0}, numInternal: {1}, totalKeysInInternal: {2}".format(numLeaf, numInternal, totalKeysInInternal))
  97. if __name__ == "__main__":
  98. # if we are being run as a real program
  100. parser = argparse.ArgumentParser(description="print out bzr index keys",
  101. epilog="Copyright Oct 14, 2014 - Mark Grandi")
  103. parser.add_argument('infileList', type=argparse.FileType('rb'), nargs="+", help="bzr index to read the keys from")
  106. printOutIndexKeys(parser.parse_args())
  107.  
Powered by spacepaste. Paste removal.